Failure to read output files - no output files in directory

877 views
Skip to first unread message

ACaragiulo

unread,
Jun 3, 2012, 12:37:41 PM6/3/12
to structure-software
Hi all,

I ran a data set of 227 diploid individuals with 21 microsat loci for
20 iterations, with K varying from 1 to 12, with a burn-in of 10,000
and a MCMC length of 50,000. It said "job complete" but there weren't
any output results files in the directory. I then noticed that it had
given warnings saying it couldn't open the results file. I tried
running the same dataset with k=1, 100 burn-in, and MCMC of 1000 and
it successfully gave me an output file. I tried a larger k, burn-in,
and MCMC, and it gave no results. Then I tried the original k=1, 100
burn-in, and 1000 MCMC to try and re-create the writing of an output
file and it wouldn't do it anymore. I also tried closing the program
and re-initiated the k=1, 100 burn-in, 1000 MCMC project and still
nothing. The STRUCTURE .exe file has full permissions (reading,
writing, etc.) on my Vista machine. Does anyone have any tips or
solutions?? Any information would be appreciated.

Thanks,
Anthony Caragiulo
Ph.D. candidate
Fordham University
American Museum of Natural History

Adii_ (Institute of Genetics and Animal Breeding)

unread,
Jun 4, 2012, 9:38:40 AM6/4/12
to structure...@googlegroups.com
If the STRUCTURE does create the output files once, it means there is no problem with reading/writing rights.

Could You tell us more about Your machine? Especially CPU and RAM :)

It will be also useful if You paste the warnings You get from STRUCTURE 

Adii_
'Models are to be used, not believed' H.Theil 

ACaragiulo

unread,
Jun 4, 2012, 2:43:51 PM6/4/12
to structure-software
Hi Adii -

Thanks for the reply. Here is the warning I get from STRUCTURE after
my run:
"WARNING: Unable to open output file C:\Users\Anthony\Documents\Fordham
\Thesis Work\Felid Project\Structure_Analysis
\Test2\Test2\10000\50000\Results\10000\50000_run_1_f."

There are no files in the Results directory either. I am running
STRUCTURE on a Dell Inspiron 1420 laptop with Windows Vista Home
Premium, Intel Core 2 Duo CPU T8300 @2.40 GHz with 4 GB of RAM. I've
also run it on two other machines running Windows XP (not sure of
their CPU and RAM) and using earlier versions of STRUCTURE (v.2.2),
and have gotten the same error message. I've also tried running the
example data that comes with STRUCTURE and I get the same error
message. The analysis is running because I can see the lnP and all
other stats scrolling in the bottom window, but there isn't an output
file.

Any help is greatly appreciated - I'm desperate to get this analysis
running for my dissertation.

Thanks,
Anthony

On Jun 4, 9:38 am, "Adii_ (Institute of Genetics and Animal Breeding)"

candice

unread,
Jun 5, 2012, 2:37:21 AM6/5/12
to structure-software
Hi Anthony,

I am running Front End on my Mac and had the same problem as you. I
was frustrated to as I'm also trying to get my PhD work done. I would
do a run and nothing would deposit in the results folder. Then I'd set
up a new set of parameters and I would get results. Next I tried a new
parameters set and again no results. To solve the problem I did the
following:
-Created a new parameter set, make sure the file folder with the
parameter set was active yet not opened.
-Go to the menu bar and select "run" from the parameter set bar (do
not use the short cut ! button)
-Process data as normal (try just one run under a short burn-in and
run length to make sure the results are created)

I'm not sure if this will work for you since you you're on Vista but I
think it's worth a shot. I'm under the impression that this sort of
error is very common with the GUI version of STRUCTURE (are you using
the GUI?) and you may be able to search this group to see if there are
additional tips on how to get your results files.

Hope this works for you.

Candice

Adii_ (Institute of Genetics and Animal Breeding)

unread,
Jun 5, 2012, 6:44:05 AM6/5/12
to structure...@googlegroups.com
Hi

Your laptops in more then enough for this analyse, so it hast to be software problem.
Many peoples with Vista or MAC got this problem. 

It seams to be the problem with path's to directories. Check in Your parameters file - it the path to directory is 100% right? 

I'm using Linux console version (Debian) and never got any problems with STRUCTURE. So i guess the computational part of STRUCTURE is very well designed and programmed. 
The problem is with user interface for Windows/MAC.

Using the Linux version will be the solution. 

I could offer help with getting started with Linux version, or I could make run some jobs on my machines. (write me an @ for details)

At the moment Your Institute is creating big cluster for outside-company users. 
I will let all know on group when I will be authorize to create accounts on our machines. 

Good luck with solving Your problem. :)

ACaragiulo

unread,
Jun 5, 2012, 9:22:08 AM6/5/12
to structure-software
Thanks Candice & Adii -

I tried STRUCTURE once more this morning and it appears to work
perfectly now. I guess it's just tempermental. I had created a new
folder and directed the results to the new folder, then ran the data
file with a new parameter set and it is writing the output file.

I checked the mainparams files for runs that didn't work and it had
added additional text to the folder pathway - maybe this was the
problem. I'm going to check the mainparams file for the successful
runs to see if it is correct. Can I change this in a text editor and
then re-run if I get the "failure to open results" error again?

Best,
Anthony

On Jun 5, 6:44 am, "Adii_ (Institute of Genetics and Animal Breeding)"

Adii_ (Institute of Genetics and Animal Breeding)

unread,
Jun 5, 2012, 9:41:32 AM6/5/12
to structure...@googlegroups.com
If there is additional text in path, (without #), this could be a reason. 

You may change the params files (both mainparams and extraparams) and re-run the jobs.
Just remember, that if You create paramfile on one OS, It may not work on other OS (f.ex. there is a problem with ^M EOL sign in files from Windows used on Linux)
So the best solution is to edit the file on the same OS. 

Good luck with Your tests :)

Adii_
'Models are to be used, not believed' H. Theil

Julie Hebert

unread,
Jun 5, 2012, 10:05:42 AM6/5/12
to structure-software
Just a general note: this is a problem many people on the forum have
had. The first thing everyone should try is to set up the input and
parameter files, start a run, stop the run, leave Structure, open it
again, then start the run again. For reasons that, as far as I can
tell, no one understands, this works. I've seen it work on XP, OS X,
Vista, and Windows 7. If you have access to Linux/Unix, the program
tends to run faster and has less bugs, but not many people do.
Good luck all!
Julie

On Jun 5, 9:41 am, "Adii_ (Institute of Genetics and Animal Breeding)"

Adii_ (Institute of Genetics and Animal Breeding)

unread,
Jun 6, 2012, 3:48:01 AM6/6/12
to structure...@googlegroups.com
I'm using the Linux console version only. I'v never had problems with running the jobs.
Of course I got some comments to the Linux version ;]

I talk with my supervisor. We'r planning to offer the free computational account on our machines in September.
All science/develpment users are welcome! 

ACaragiulo

unread,
Jun 6, 2012, 9:23:24 AM6/6/12
to structure-software
Thanks everyone. It seems to be working with everyone's suggestions.

On Jun 6, 3:48 am, "Adii_ (Institute of Genetics and Animal Breeding)"

Dyanna L.

unread,
Jun 14, 2012, 1:22:41 PM6/14/12
to structure...@googlegroups.com
Hi Candice,

I'm also having trouble on my Mac. I tried your advice above with no success. Could you tell me what version of Structure you are using and what OS you have in your Mac?

Thanks, 

Dyanna

Berthold

unread,
May 3, 2013, 5:38:23 AM5/3/13
to structure...@googlegroups.com
Hello all

I had a similar problem with STRUCTURE not saving my results from long batch jobs. It seemed to have appeared very recently after a JAVA update on my computer, but ....

After reading all your hints, I checked my mainparams file and saw the pretty long file pathes - more than 200 characters in one case. I tried again with the same data file, new project, same parameters, but a much shorter path (approx. 120 characters), and a run with less iterations has worked just now. So my impression is that __long pathes and filenames__ may be the problem.

best regards,
Berthold

Berthold Heinze

unread,
May 4, 2013, 1:11:04 PM5/4/13
to structure...@googlegroups.com
Hello again,

I solved my problem (STRUCTURE not saving results after several runs in a batch job). It wasn't the updated Java, it wasn't long pathes, it was simply that the network drive where I store data was completely full ... I can't imagine it could happen to anybody else, but who knows ... thought you might want to hear ...

best regards,
Berthold



--
You received this message because you are subscribed to the Google Groups "structure-software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to structure-softw...@googlegroups.com.
To post to this group, send email to structure...@googlegroups.com.
Visit this group at http://groups.google.com/group/structure-software?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Reply all
Reply to author
Forward
0 new messages